Our laboratories have been working with the CVB since the early 1980s to understand how the viruses induce human inflammatory heart disease (myocarditis). It was therefore a natural extension of our work to examine the putative connection between CVB infection and type 1 diabetes (T1D) onset.
We use the nonobese diabetic (NOD) mouse as the animal model in which to study T1D onset. This is a well-established model used throughout the world for T1D research and one which is very useful for studying aspects of the virus-host relationship. Female NOD mice develop T1D at an incidence of between 70-100% of mice by 6 months of age: this means that for every 10 mice studied, 7-10 will naturally develop T1D by 6 months of age. Examination of the pancreatic islets in these mice shows that when mice are very young, no insulitis is apparent but by 6-8 weeks of age, insulitis has started to develop. Insulitis is inflammation of the islets, the places in the pancreas where beta cells are found. Beta cells produce insulin. When enough beta cells are destroyed, T1D occurs. Islet inflammation is autoimmune, which is to say, it is a naturally occurring inflammation that targets the host itself. By 12-15 weeks, insulitis is extensive in nearly every islet and it is at this age that the mice begin to develop outright T1D. This is easily observed by measuring the level of glucose (sugar) in the mouse' urine: when normal, there is no glucose detectable but once diabetic, the mice shed more than 20 grams per liter of urine (equal to about an ounce per quart).
(A picture of a NOD mouse pancreatic islet that is dying due to the infiltration of autoimmune lymphocytes. The pathogenic lymphocytes are the dark cells surrounding the interior, lighter area, which is the remaining intact islet, still able to produce insulin. But not for long...)
Our laboratory has a collection of different CVB strains for its studies. A serotype of CVB classifies a group of CVB; we use predominantly the CVB3 because we have spent most of our research characterizing this specific serotype. However, many different enteroviruses are likely able to cause T1D in humans, not just the CVB. It is often mentioned that only CVB4 causes T1D: this is simply not true. Now, within any CVB serotype, there are numerous strains of viruses, which all differ genetically from each other. You can think of a virus strain as a variation on a single theme. Our use of the CVB3 strains has permitted a deeper understanding of how relatively minor variations in the viral genetics can have huge impact on the outcomes of virus infections. We have derived molecular clones of several CVB3 genomes so that we can manipulate the viral genetic stuff in the test tube, then resurrect infectious virus in cell cultures, and use such viruses to study their biologies. This is a technique called reverse genetics and uses another key technique, molecular cloning. We have discovered several new aspects of the virus/T1D relationship using the mouse model, all of which are consistent with that which is known from others' human studies.
1. The CVB protect diabetes prone mice from developing T1D.
Because the CVB are often mentioned as primary infectious causes of human T1D, we asked the simple question: if we inoculate the virus into young, healthy NOD mice, what happens? Does T1D immediately occur? The answer? No! Such CVB-inoculated mice enjoyed a significantly diminished chance of developing T1D compared to control mice (mice which were not inoculated with virus and develop T1D normally). In some cases following CVB inoculation, no mice developed T1D through 10 months of life. This finding showed that there is no simple link between these viruses and T1D. The NOD mouse is very prone to developing T1D: these data showed, however, that a common virus infection, one linked to human T1D onset, could actually protect these mice.
There is the criticism that nearly any treatment of NOD mice will suppress T1D and in large part, this is true. However, this criticism ignores the important fact that alone of all the treatments experimentally used in NOD mice to suppress T1D, inoculation with CVB represents a test of an agent suspected to be the cause - not the cure - of T1D. Our work demonstrated that, in effect, we can vaccinate NOD mice so that they do not develop T1D. This in turn suggests the intriguing possibility that one might be able to be vaccinated against developing T1D. Indeed, we strongly believe that T1D was rare in humans before about 100-200 years ago, simply because humans were commonly exposed naturally to numerous enterovirus infections as a natural part of growing up in a world of contaminated water and poor or absent hygiene . [This was recently reviewed: Enteroviruses, type 1 diabetes, and hygiene: a complex relationship. S. Tracy et al., Reviews in Medical Virology 20:106-116, 2010.]
2. The CVB do not invade and destroy islet cells of healthy mice.
Viruses generally destroy cells by direct infection: viruses enter a cell, take it over, replicate themselves and in the process, kill the cell, releasing newly-created progeny virus to repeat this process. If enteroviruses such as the CVB are to be considered causes of T1D, then - most simply - the viruses must be able to destroy the insulin-producing beta cells in the pancreatic Islets of Langerhans. We observed that no virus was detectable within the islets of young, healthy NOD mice, even though we could detect the receptor protein that the CVB uses to gain entrance to cells. Receptors are like doors to rooms: a virus has to have a receptor in order to gain entrance to a host cell. Thus, even though we showed the receptor is present in islets, the 'door' appears somehow barred to effective CVB entry. This observation was consistent with our failure to observe that CVB cause T1D in young, healthy mice: if the virus cannot kill islet cells, then one would suspect the virus cannot induce T1D, either. In fact, this is what we observed.
(This is a picture of a human pancreatic islet that was stained for the expression of a protein, called CAR, the receptor which the coxsackie B viruses require in order to enter a cell to replicate. The dark brown is the islet to which an antibody against CAR is bound. Clearly, human islets, like mouse islets, express CAR and so, should be able to be infected under the right circumstances.)
3. However, if the islet microenvironment is altered in specific ways, CVB can enter the islets.
Other workers have suggested that the islets defend themselves against virus entry through a mechanism called the innate immune response and production of specific antiviral protein molecules called interferons. Using a CVB3 strain that we developed in the laboratory which was bioengineered to produce a mouse immune protein (cytokine) called interleukin-4 (or IL-4), we showed that this virus did gain entry to islets in young, healthy NOD mice. This experiment was important for two reasons. One, it showed that the expression of the virus receptor meant that CVB could gain access to islet cells. While this was logical, it had not been shown before in the mouse itself, only in a special condition (cell culture). Secondly, it showed that by changing the local microenvironment of the islet by the virus-induced production of IL-4, the virus could replicate successfully in the islets.
We also noticed two more things of importance. One, this type of virus infection caused no insulitis: the virus which produced IL-4 did not induce the mouse to attack the islets with its anti-viral immune response. Two, mice inoculated with this strain of virus had a better chance of never developing T1D than mice which did not get the virus infection. This surprising finding meant that despite intraislet replication of this bioengineered virus, this group of mice developed fewer cases of T1D than did mice without the virus injection. This observation showed that in some cases, virus infection of islets does not lead to more T1D or rapid onset T1D and therefore, the story was not quite so simple .
Islets in older NOD mice naturally become massively inflamed with autoimmune lymphocytes. This kills beta cells and the islet then loses the ability to produce insulin. Here you can see the residual small areas in such an islet in which coxsackie B virus is replicating (shown by the brown color in the upper left corner primarily). The light blue in the remaining area are pathogenic autoimmune lymphocytes. Virus only replicates in the remaining healthy tissue, thus speeding T1D onset by killing insulin-producing beta cells.
4. CVB infection of older, prediabetic mice can, however, trigger T1D, an event linked to the alteration of the islet microenvironment.
Older, pre-diabetic mice show massive insulitis in nearly every islet and not surprisingly, soon begin to become sick with T1D due to loss of insulin production. This naturally occurring, genetically driven autoimmune disease kills cells in the islets, including the beta cells, a process that leads to loss of insulin production and thus, T1D onset.
This inflammation of islets represents a real change in the biology of the islet, a massive naturally-occurring change in the islet microenvironment. We therefore asked another simple question: if CVB does not trigger T1D in young mice but instead, protects them, what happens in mice that are about to develop T1D anyway? We knew that IL-4 could let virus replicate in islets: would inflammation permit the same? The answer was yes!
Young NOD mice are analogous to humans who are genetically predisposed to developing autoimmune T1D but have yet to do so: they may have little or no insulitis present, just like young NOD mice. Based on our results, we suggest that humans with little or no insulitis, are at low risk from CVB-induced T1D. This is because we have shown that CVB (or in humans, we believe other enteroviruses as well) need to have insulitis in place in order to be able to successfully replicate in islets. However, it is quite difficult to say how advanced insulitis is in a human being; even presuming one knows that one is at risk. Humans are not like mice in a key respect: these mice are highly inbred and so, their own genes drive them to develop T1D in a regular, predictable fashion (the very thing that makes them so useful for this research). Every human is genetically distinct and has a different schedule for developing autoimmune insulitis (if indeed they ever do and of course, by far most do not). By modeling this situation in mice, we mimic the case in humans where a virus infection occurs at a time closely prior to the time when that person would develop T1D anyway from his/her own autoimmune disease. In mice, this is a specific age; in humans, it could be any time.
What we found was that pre-diabetic mice - i.e., mice with ongoing insulitis - when inoculated with a virulent strain of CVB, rapidly developed T1D, much faster than the rate of development observed in the control mice in which it is controlled only by the autoimmune disease. When we examined the islets of such mice, we discovered the presence of virus (as shown above). That virus was found replicating within the islets and associated with beta cells prior to the onset of T1D meant that the virus replication was denuding the mouse of intact beta cells, consequently causing early onset T1D.
(Stained bright red is an isolated mouse pancreatic islet. It is still associated with some residual pancreatic tissue.)
5. Findings in mice and how they relate to human T1D: connecting the dots.
The very great majority of enterovirus infections in humans never trigger T1D, even though enterovirus infections are common in the US from spring through the summer into the fall months and enteroviruses are encountered worldwide. So, if human enteroviruses are causes of human T1D, how is this explained? Using information we have gained by asking key questions of our mouse model and correlating clinical reports of enterovirus-linked T1D, we suggest that there are several reasons.
It is very likely that only certain enteroviruses can induce T1D or for that matter, act to protect one from developing T1D. Clearly in NOD mice, the CVB can either protect mice from developing autoimmune T1D (when they are exposed to the virus when young) or CVB can rapidly trigger T1D onset (when older mice with insulitis are exposed to the virus). The CVB belong to one of four human enterovirus species, denoted A-D. Only human enteroviruses of the B species (or HEV-B), and this includes the CVB, have been associated with T1D onset. Poliovirus, for example, which is a species C enterovirus and to which nearly everyone in the world has been exposed (mostly now by clinical inoculations but prior to this due to wild-type infections), has had no impact on T1D incidence. Thus, we propose that the HEV-B species are the key players.
Now, for an enterovirus to 'suddenly' trigger T1D (when T1D occurs shortly after or during, for example, a 'cold' or 'flu-like' illness), we believe the islets have to already be significantly experiencing extensive insulitis through one's own autoimmune disease. That is to say, insulitis has to be present. This might normally, in time, lead to T1D onset or it might not. From the NOD mouse model, we know that islets in young mice that are not inflamed cannot be normally infected by CVB. This does not have to do with the virus receptor, the protein on the cell surface which the virus uses to enter cells. The CVB receptor is well expressed in young and older mice. So, when mice are young and have no insulitis yet, the islets cannot be infected, but when the mice are older and are developing insulitis, CVB can infect remaining healthy islet tissue and if the viral damage is sufficient, T1D ensues shortly after the virus infection. However, the overwhelming majority of people do not have insulitis. Therefore, we postulate that because most enterovirus infections in humans do not induce T1D, by far most people do not have insulitis. Therefore, this is consistent with the observation that the very great majority of enterovirus infections do not trigger T1D onset.
The host (humans or mice) have to "work" with the virus to cause T1D. That is to say, without host-driven (genetically determined) insulitis, the enterovirus cannot replicate productively in beta cells and cause T1D. Viruses are opportunists and will replicate wherever they can. In the case of a normal (not inflamed) islet in the mouse pancreas, CVB can enter cells (because the receptor is present) but cannot successfully replicate. We hypothesize that this is also the case in human beings. Only when the islet is attacked by the host's autoimmune disease do islets' defenses fall, permitting the virus to replicate productively in and kill islet cells. We know this happens in mice and we postulate this is the case in human beings. But this is a contested point. Human islets, isolated from pancreas and placed in culture, can replicate enteroviruses: such infections can kill beta cells in these cultured islets. This observation suggests that human islets might be infectable in the body, whether or not they are inflamed due to the autoimmune process. Currently, this remains an open question.
We also know from our work in mice, that the enterovirus infection has to be due to a strain that replicates quickly in the pancreas. Just like some humans can run faster than others, some virus strains can replicate faster than others (that is, some virus strains make more progeny virus in a shorter length of time than others). We have characterized strains (or variants) of CVB serotypes that replicate more rapidly and to higher titers, than other strains. To initiate T1D in NOD mice, we have shown that as few as 50 virus particles of a rapidly replicating strain of CVB3 can induce T1D, whereas more than 1 million virus particles of a slowly replicating strain are needed to induce T1D. Therefore, the average dose needed to successfully infect a mouse and to cause T1D with a rapidly replicating strain is far lower than for other strains. However, these rapidly replicating strains of enterovirus (which are generally termed 'virulent' strains, due to their capacity to induce disease easily in disease models) which are capable of causing severe disease, circulate relatively rarely. Most CVB strains, for example, do not cause serious disease such as myocarditis when assayed in mice. And even though all CVB strains replicate well in pancreas (thereby showing that these viruses have a predilection for replicating in pancreas tissue), this does not mean, that all are capable of replicating sufficiently well in islets to trigger T1D. The bottom line is this: the average (usual) enterovirus infection is due to a poorly virulent strain at a low dose, two factors that along with the requirement for ongoing insulitis, lower the odds dramatically for a T1D-inducing islet infection.
We also know that enterovirus infections induce protective antiviral immunity in people. This is the same principle by which the poliovirus vaccines have worked so well: inoculation with the vaccine strains of poliovirus induce an immunity that dramatically suppresses the replication of polioviruses when the human again is infected, thereby keeping that individual safe from crippling polio. There are more than 100 known enterovirus serotypes and each serotype induces immunity in a person that will protect that person from disease caused by future exposure to that same serotype. This means that if one has already experienced an infection by a specific serotype, for example CVB3, one is immune to disease from all variations (strains) of CVB3 when next one may encounter it. However, type-specific immunity does not protect one from infection by a different serotype; protection is serotype-specific. To continue this example, therefore, CVB3 immunity will not protect one against infection by a strain of CVB1 or CVB4, for instance. Therefore, in order to trigger T1D in humans, an ebterovirus must infect a person who has no pre-existing immunity to that specific virus. So in addition to everything else discussed above, one must also experience a new enterovirus infection, against which one has no immunity, in order for T1D to be triggered. From this argument -knowing the various requirements which we can postulate to exist based on our current knowledge - one can see that having T1D initiated by a enterovirus infection such that it 'suddenly' occurs, would be a rare event.
In order for T1D to be triggered by a enterovirus infection, therefore, a variety of specific conditions have to be met all at the same time: (1) the right enterovirus species (not all can do this, insulitis needs to be present (and most people likely have none), (2) the virus strain should be one that replicates rapidly (because the average natural infectious dose is very low, the virus has to generate enough progeny virus to cause the damage before the host immune response suppresses the infection), (3) the virus infection must be one never before encountered by the person (otherwise, that person is immune to the virus), and (4) the person's islets must have insulitis ongoing (in order to create the environment that supports productive enterovirus replication in the islets). If T1D onset triggered by an enterovirus requires all these requirements, one can understand why the disease is rarely caused by an enterovirus.
6. What does all of this mean for a cure for T1D?
Finding a cure for existing T1D or a preventive measure against as-yet-to-occur T1D are two vital missions. Current work suggests that newly diagnosed T1D patients may profit from an antibody treatment that reduces pathogenic T cells, permitting the patient's own regulatory (good) T cells to expand in number to protect the islets from damage. This is wonderful news. However, we must also stay focused on the issue of preventing T1D completely. We know the value of vaccines: polio, rabies, measles, mumps, rubella and more, all are diseases readily countered and suppressed by vaccine development. Properly designed vaccines work. However, the relationship between enterovirus infections and T1D biology is complex as the foregoing arguments have shown.
We can largely prevent T1D in NOD mice with a single injection of CVB at an early age. This means we can prevent the host's own autoimmune disease from killing the beta cells and causing T1D - in most cases. There are those who argue that NOD mice are not a good platform for designing approaches to counteract or suppress T1D, and in most cases, this criticism is valid: nearly every approach that has functioned well in NOD mice does not function in humans. However, human enteroviruses are human viruses and we know they are involved in human T1D. That they mimic much of what we know or surmise occurs when studied in the NOD mouse, is strongly inferential data in support of the hypothesis that certain human enteroviruses can either protect from, or induce, T1D. Enteroviruses are 'a bird in the hand' argument: we know they are involved in human T1D. So while we wait for clinical studies to produce lists of potential infectious candidates involved in the T1D etiology, we ought to be moving ahead to understand how enteroviruses are involved in the disease process. In the final analysis, no matter what list of potential pathogens that are found to be suspected of causing T1D, the human enteroviruses will be at the top of the list. Waiting is not an option anymore, now that we know this story.
Can a vaccine be created against enteroviruses which will eliminate T1D and perhaps all other enteroviral diseases as well? Good question. Vaccines usually target at most a few viruses; the polio vaccines targeted three types of poliovirus, for example. A major developmental problem for T1D is that at present, we do not know which enteroviruses cause T1D: there are at least 100 known and many more uncharacterized human enteroviruses. This is far too many to target for a vaccine, especially if some or many play no role in the disease. For a classical vaccine approach to be considered, we must determine which enteroviruses cause T1D, and determine how the viruses cause the disease. That enteroviruses in species B are likely the key players, suggests the relevant field has been winnowed significantly already. Research on this topic should be actively encouraged.
Of course, there is also the question of how many cases of T1D are indeed caused by enteroviral infections. If enteroviruses do not cause many cases, no company would ever make a vaccine because the market would be so small. That is a hard fact: cures are dependent upon the free market. We know that CVB and other species B enteroviruses are involved in T1D induction. However, work is required to identify the viruses that are found in the pancreas tissue of diabetic humans.
That said, we have been speaking so far about a classical vaccine: one which develops protective antiviral immunity against specific virus(es) serotypes that protects one against disease caused by subsequent exposure to the same virus(es) serotypes. There are potentially other approaches to vaccination, ones which may not involve the generation of protective immunity but instead, a generic or pan-enterovirus immunity. The possibility exists that one can induce the immune system to recognize enteroviruses in general, such that when it comes up against an actual virus infection, the immune response will be much more rapid. That may be all that is required and that would be rather readily accomplished. Such an approach is not a "silver bullet" like, for example, the polio vaccines; such an approach would offer a much better chance at suppressing a potentially T1D-causing infection but might protect all people. This, too, is worthy of research support because of its immense potential, not only for T1D but for viral diseases in general. In fact, we now have a large amount of data demonstrating that we can actually vaccinate NOD mice in this way and protect them not only from their own autoimmune T1D but also from CVB-induced T1D later in life. Therefore, we believe that the potential for this approach is huge. In one scenario, using standard and well-understood (from the poliovirus vaccine experience) technologies, a safe, protective anti-enteroviral vaccine could be devised. While it would not completely protect the individual from virus disease in the same manner that the polio vaccines prevent poliomyelitis, such an approach could slow the infection sufficiently so that the immune response would have a vital few extra days to respond and clear the infection. Again, data from the mouse model indicates this slowing of new infections is tightly linked to lowering the chance of developing virus-induced T1D.
7. The importance of basic research to help find the cure to T1D.
That which we understand about human enteroviruses and their impact upon T1D development has been derived from basic research. As scientists and citizens, we are driven by the need for a cure, but we must temper our approach as we know that if we chase off down a promising but blind alley, we will waste valuable time. Ignoring basic research while emphasizing only clinical palliative efforts will never eradicate or suppress the disease. The goal should always be to eradicate the disease, while an acceptable compromise is to greatly reduce the incidence of the disease. To focus primarily on treating people with the disease is unacceptable.